Inferring Regulatory Networks from Expression Data Using Tree-Based Methods

By Vân Anh Huynh-Thu, A. Irrthum, Louis Wehenkel, Pierre Geurts. 2010

2,004 citations recorded by the source corpus.

One of the pressing open problems of computational systems biology is the elucidation of the topology of genetic regulatory networks (GRNs) using high throughput genomic data, in particular microarray gene expression data. The Dialogue for Reverse Engineering Assessments and Methods (DREAM) challenge aims to evaluate the success of GRN inference algorithms on benchmarks of simulated data. In this article, we present GENIE3, a new algorithm for the inference of GRNs that was best performer in the DREAM4 In Silico Multifactorial challenge. GENIE3 decomposes the prediction of a regulatory network between p genes into p different regression problems. In each of the regression problems, the expression pattern of one of the genes (target gene) is predicted from the expression patterns of all the other genes (input genes), using tree-based ensemble methods Random Forests or Extra-Trees. The importance of an input gene in the prediction of the target gene expression pattern is taken as an indication of a putative regulatory link. Putative regulatory links are then aggregated over all genes to provide a ranking of interactions from which the whole network is reconstructed. In addition to performing well on the DREAM4 In Silico Multifactorial challenge simulated data, we show that GENIE3 compares favorably with existing algorithms to decipher the genetic regulatory network of Escherichia coli. It doesn't make any assumption about the nature of gene regulation, can deal with combinatorial and non-linear interactions, produces directed GRNs, and is fast and scalable. In conclusion, we propose a new algorithm for GRN inference that performs well on both synthetic and real gene expression data. The algorithm, based on feature selection with tree-based ensemble methods, is simple and generic, making it adaptable to other types of genomic data and interactions.

Subfields: Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

Citations in the corpus

Consistent Individualized Feature Attribution for Tree Ensembles (2018, neutral): “Though its motivation is largely heuristic [11], gain is widely used as the basis for feature selection methods [5, 13, 25].”
SCENIC: Single-cell regulatory network inference and clustering (2017, methodology): “To achieve high-quality modules, we combined existing methods for TF-target inference that were originally developed for bulk RNA-seq or microarray data, namely GENIE3 32 and gradient boosting 72 , and found that they perform adequately on single-cell RNA-seq data.”
PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data (2019, methodology): “We also explored creating co-expression networks using GENIE3 (26) and WGCNA (27), the former was deemed too computationally expensive and the latter did not work well due to the sparse properties of the single cell sequencing data.”
TIGRESS: Trustful Inference of Gene REgulation using Stability Selection (2012, methodology): “We now propose a new scoring function sg(t) to assess the significance of a transcription factor t ∈ Tg in the regression model (1).”
Understanding Random Forests: From Theory to Practice (2014, methodology): “n factors on the basis of their expression level, in order to reconstruct a global network of interactions (as illustrated in Figure 7.9 for Staphylococcus aureus). As proposed in the GENIE 3 method [Huynh-Thu et al., 2010 ], the network inference problem can be solved generically by remarking that ”
Wisdom of crowds for robust gene network inference (2012, methodology): “The two teams with the best overall score used novel inference approaches based on random forests 19 and ANOVA 20 (Other 1 and 2), respectively (Table 1).”
SCODE: an efficient regulatory network inference algorithm from single-cell RNA-Seq during differentiation (2016, methodology): “In addition, we inferred networks with GENIE3 [26], which also predicts TF expression from the expression of other TFs by using regression trees.”
Detecting gene-gene interactions using a permutation-based random forest method (2016, background): “People have successfully developed tree-based meth-ods to infer gene regulatory networks [19].”
Data Based Identification and Prediction of Nonlinear and Complex Dynamical Systems (2016, methodology): “A number of methods were developed to accomplish this goal, with some works showing that the performance of the supervised methods is better than that of the unsupervised methods [357].”
Structural Agnostic Modeling: Adversarial Learning of Causal Graphs (2018, methodology): “ted with the sparsebn package (Aragam et al., 2017). 10. The better-performing, order-independent version of the PC algorithm proposed by Colombo and Maathuis (2014) is used. 18 The GENIE3 algorithm (Irrthum et al., 2010) is also considered, though it does not focus on DAG discovery per se as it ach”
ScholarWorks @ UTRGV ScholarWorks @ UTRGV (methodology): “After genes and SNPs were selected from related Genome-Wide Association Study (GWAS), it was tested how the method identify true positive edges between genes and SNPs without eQTL information.”